llvm-project

[AArch64] Lower i16/f16 bitcast via vector operations.

#196341

Merged